¹ The literature bearing on these points is extensive. It includes Lazarsfeld, Paul F., Berelson, Bernard, and Gaudet, Hazel, The People's Choice: How the Voter Makes Up His Mind in a Presidential Campaign (New York: Columbia University Press, 1944)Google Scholar, which emphasizes social factors in voting; Berelson, Bernard, Lazarsfeld, Paul F., and McPhee, William N., Voting: A Study of Opinion Formation in a Presidential Campaign (Chicago: University of Chicago Press, 1954)Google Scholar, which explores social influences on the vote, and misperceptions of candidates' issue positions; Campbell, Angus, Converse, Philip E., Miller, Warren E., and Stokes, Donald E., The American Voter (New York: Wiley, 1960)Google Scholar, which establishes the primacy of party identification, and shows the weakness of opinions and perceptions about issues; Converse, Philip E., “The Nature of Belief Systems in Mass Publics,” in Ideology and Discontent ed. Apter, David E. (New York: Free Press, 1964), pp. 206–261Google Scholar, which demonstrates that opinions on policy lacked any simple ideological structure, and were unstable over a four-year period; and Stokes, Donald E., “Some Dynamic Elements of Contests for the Presidency,” American Political Science Review, 60 (03 1966), 19–28CrossRefGoogle Scholar, which shows that issues have been less important than candidate images as “short term forces” affecting electoral outcomes.

The late V. O. Key manfully tried to find evidence of an issue-oriented “responsible electorate” among those who switched their votes from one election to the next, but his effort was beset by methodological difficulties. Respondents' recollections of their votes four years previously, which Key used to identify switchers, are not very reliable; the issues upon which he found agreement between opinions and direction of switch were broadly phrased judgments about governmental performance, which are particularly subject to rationalization by voters who switch for other reasons. Key, V. O. Jr., The Responsible Electorate (Cambridge, Mass.: Harvard University Press, 1966)CrossRefGoogle Scholar.

² Key, V. O. Jr., Public Opinion and American Democracy (New York: Knopf, 1961), pp. 458–480Google Scholar. Kramer, Gerald H., “Short-Term Fluctuations in U.S. Voting Behavior, 1896–1964,” American Political Science Review, 65 (03 1971), 131–143CrossRefGoogle Scholar, provides some evidence of electoral punishment in response to bad economic conditions.

³ The existence of “issue publics” is demonstrated in Converse, “Belief Systems,” pp. 245–246. RePass, David E., “Issue Salience and Party Choice,” American Political Science Review, 65 (06 1971), 389–400CrossRefGoogle Scholar, finds a strong correspondence between voting choices and the party whose stand is preferred on salient issues; as he points out, however (p. 395), some of this correspondence may result from rationalization.

⁴ Electoral competition models which predict party convergence are set forth in Hotelling, Harold, “Stability in Competition,” Economic Journal, 39 (03 1929), 41–57CrossRefGoogle Scholar; Downs, Anthony, An Economic Theory of Democracy (New York: Harper, 1957), pp. 114–141Google Scholar; and Davis, Otto A., Hinich, Melvin J., and Ordeshook, Peter C., “An Expository Development of a Mathematical Model of the Electoral Process,” American Political Science Review, 64 (06 1970), 426–448CrossRefGoogle Scholar.

⁵ Kelley, Stanley, Political Campaigning (Washington, D.C.: Brookings, 1960)Google Scholar finds considerable evidence of candidate ambiguity. Theoretical arguments predicting ambiguity are given in Downs, pp. 135–137; and in Shepsle, Kenneth A., “The Strategy of Ambiguity: Uncertainty and Electoral Competition,” American Political Science Review, 66 (06, 1972), pp. 555–569CrossRefGoogle Scholar.

⁶ In the IEP project as a whole, we are studying a variety of questions about the issue-related behavior of citizens and candidates, including the other challenges mentioned above.

⁷ Gallup Opinion Index, 04 1968, p. 3, 14Google Scholar.

⁸ The surveys were conducted in February, June, August, and November 1968 by the Opinion Research Corporation (Princeton, New Jersey) for the IEP project. In addition to the present authors, the project includes Jerome Laulicht and Sidney Verba.

⁹ In 1967, specific Vietnam items formed a satisfactory Guttman scale, with CR = .88 and CS = .66. Richard A. Brody and Sidney Verba, “Hawk and Dove: The Search for an Explanation of Vietnam Policy Preferences,” Acta Politica, forthcoming. In an earlier paper, based on data from 1966, it was reported that opinions fell into two imperfectly related scales, one expressing attitudes toward escalation, and the other attitudes toward de-escalation. Verba, Sidney, Brody, Richard A., Parker, Edwin B., Nie, Norman H., Polsby, Nelson W., Ekman, Paul, and Black, Gordon S., “Public Opinion and the War in Vietnam,” American Political Science Review, 61 (06 1967), 317–333CrossRefGoogle Scholar, material cited is on pp. 320–321. Reinterpretation of those data suggests that a single escalation/de-escalation dimension might have existed even as early as 1966.

¹⁰ In 1968 we did not have enough policy items to replicate the Guttman scaling analyses of 1966 and 1967. We found, however, that opinions on the Vietnam rating scale correlated quite highly with opinions concerning bombing, troop levels, and three general policy alternatives.

¹¹ In the SRC survey, evaluations were measured on a 100-point “thermometer” which yielded almost precisely the same levels of evaluation and the same relationships with other variables as the five-point ORC/IEP favorability scale. Page, Benjamin I., “Party Loyalty and the Popularity of Presidential Candidates” (unpublished ditto, 02., 1972)Google Scholar. For a discussion of the candidate thermometer, and some findings based on it, see Weisberg, Herbert F. and Rusk, Jerrold G., “Dimensions of Candidate Evaluation,” American Political Science Review, 64 (December 1970), 1167–1185CrossRefGoogle Scholar.

¹² The approach to policy voting which is sketched here would blend two theoretical and two methodological traditions. A formal model of issue voting drawn from the electoral competition literature (see Davis et al.) would be supplemented by psychologists' understanding of the processes of projection and persuasion. Simultaneous equation techniques from econometrics would be combined with Donald Stokes's method of computing net effects from means and estimated slopes. (See Stokes, , “Some Dynamic Elements of Contests for the Presidency,” pp. 27–28.Google Scholar) See Brody, Richard A. and Page, Benjamin I., “Comment: The Assessment of Policy Voting,” American Political Science Review, 66 (06, 1972), pp. 450–459CrossRefGoogle Scholar.

Our early efforts to perform such an analysis with respect to the Vietnam issue are reported in Brody et al., “Vietnam, the Urban Crisis and the 1968 Presidential Election” and in Brody and Page, “Issues in an Election.” It will not be possible to reach a precise conclusion about the impact of the issue without completing estimation of a nonlinear three-equation model. As is argued below, however, such a degree of precision is not needed for our present purpose.

Empirical work with perceived issue differences (or “distances,” or “proximities”), in addition to our own, includes Shapiro, Michael J., “Rational Political Man: A Synthesis of Economic and Social-Psychological Perspectives,” American Political Science Review, 63 (December 1969), 1106–1119CrossRefGoogle Scholar; and papers by David M. Kovenock and James W. Prothro of the Comparative State Elections project.

¹³ The traditional technique has one great advantage over an analysis which uses perceived issue differences: the opinion-vote relationship is not inflated by projection, in which respondents tend to imagine that the positions of favored candidates are the same as their own opinions. It has two major defects, however: (1) It may understate the importance of policy voting as a psychological process among individuals. Two people with the same policy preferences can receive different information (even different but correct information) about the stands of candidates; even if they both vote in accord with their opinions and their information about candidates' positions, they may vote for different candidates and thereby reduce the apparent relationship between opinion and vote. As we argue above, precise assessment of policy voting requires explicit consideration of perceptions. (2) It may overstate the amount of policy voting: some people whose opinions and votes correspond may have brought opinions into line with intended vote, rather than vice versa.

In most cases the second defect is probably the more serious. Since it inflates the relationship found in the table, however, it cannot impeach a finding of little or no relationship. The first defect, which can lead to an underestimate of policy voting, requires that even a claim of no relationship be qualified, or that it be supported by additional evidence.

¹⁴ This claim requires evidence that the weak relationship in the table does not conceal policy voting by those with differing perceptions. In previous work, employing the concept of issue distance, we also found the effects of Vietnam policy preferences to be relatively small. Brody et al., “Vietnam, the Urban Crisis, and the 1968 Presidential Election;” Brody and Page, “Issues in an Election.” Since those findings inflated the apparent impact of Vietnam, by accepting all opinions and perceptions as causally prior to the vote, they support the present finding a fortiori. Further evidence is presented in the next section.

¹⁵ According to the coefficient of determination (r²), in a linear regression, variations in opinion could account for only 1.0 per cent or 1.2 per cent of the variance in voting. According to the correlation ratio (eta), which better reflects the nonlinear relationship, variations in preference could account for 1.5 per cent of 1.8 per cent of the variance in the vote.

¹⁶ This conclusion must be distinguished sharply from the assertion that Vietnam opinions had little effect upon the 1968 election. We are concerned here only with preferences about future policy, not with evaluations of past performance in Vietnam. Further, we are concerned only with the effects of these attitudes on voting choices between Humphrey and Nixon; we do not consider their effects on voting for Wallace, abstentions from voting, or decisions whether or not to give time and money to the campaign, to say nothing of effects on the launching and subsequent fate of the prenomination candidacies of Eugene McCarthy and Robert Kennedy, or the withdrawal from candidacy of President Johnson. From this broader perspective the impact of Vietnam opinions seems to have been profound.

¹⁷ This difference is statistically significant, according to a one tailed t-test of the means, but substantively it is quite small.

¹⁸ Fifty-seven per cent saw little or no difference. Thirty per cent saw no difference at all: they placed Nixon and Humphrey at exactly the same point on the scale. Similarly, in response to another SRC item, 33 per cent replied that they saw no difference between the two parties with respect to three general Vietnam policy alternatives.

¹⁹ These people did not in fact vote their Vietnam preferences. Among them, the relationship between opinion and vote was irregular.

²⁰ We do not assume that the candidates' speeches reflected their true beliefs or intentions. They did, however, represent the best information about those intentions which was available to the public.

²¹ Humphrey's speeches were kindly provided by Raymond Wolfinger, John Stewart, and the Democratic National Committee; Nixon's were provided by Representative Charles Gubser and the Republican National Committee. The texts of some of Nixon's speeches are reprinted in Nixon Speaks Out (New York: Nixon-Agnew Campaign Committee, 1968)Google Scholar.

We have also used the New York Times (hereafter NYT) to find comments which candidates made in interviews, question and answer sessions, and the like.

²² A transcript of this speech is printed in NYT, October 1, 1968, p. 33. Our citations refer to the press release.

²³ “Vietnam,” Statement to the Committee on Resolutions, August 1, 1968.

²⁴ Portions of Nixon's answers are given in Nixon on the Issues (New York: Nixon-Agnew Campaign Committee. 1968), pp. 6–12Google Scholar.

²⁵ Humphrey was consistent after September 30; prior to his Salt Lake City speech he had vacillated, with a tendency to be more hawkish.

²⁶ “Vietnam,” August 1, 1968.

²⁷ Television interview, Dallas, October 11, 1968, p. 13. This and subsequent page numbers refer—unless otherwise noted—to official press releases and transcripts.

²⁸ UPI Conference, October 7, 1968, in Nixon on the Issues, p. 6.

²⁹ Television interview, Dallas, October 11, 1968, p. 13.

³⁰ “Vietnam,” August 1, 1968.

³¹ Statement, Miami, September 30, 1968, p. 1.

³² NYT, September 26, 1968, p. 1.

³³ Television interview, Seattle, September 24, 1968, p. 2.

³⁴ Remarks to Liberal Party Executive Committee, New York, August 17, 1968, p. 2.

³⁵ NYT, July 8, 1968, p. 31.

³⁶ Remarks to Liberal Party, p. 2.

³⁷ Remarks, Toledo, September 23, 1968, p. 3.

³⁸ Televised address, Salt Lake City, September 30, 1968, p. 5.

³⁹ Salt Lake City address, p. 6.

⁴⁰ NYT, October 18, 1968, p. 35.

⁴¹ UPI Conference, p. 9.

⁴² UPI Conference, p. 9.

⁴³ Salt Lake City address, p. 5.

⁴⁴ Salt Lake City address, p. 5.

⁴⁵ NYT, October 2, 1968, p. 1, 27.

⁴⁶ It is possible that this difference resulted from greater knowledge Humphrey may have had about administration intentions. The bombing was stopped shortly before election day.

⁴⁷ UPI Conference, p. 7.

⁴⁸ UPI Conference, p. 7.

⁴⁹ Remarks to Liberal Party, p. 1.

⁵⁰ Remarks to Liberal Party, p. 2.

⁵¹ Salt Lake City address, p. 6.

⁵² NYT, July 20, 1968, p. 12.

⁵³ Hotelling, “Stability in Competition”; Downs, An Economic Theory of Democracy; Davis, “Mathematical Model of the Electoral Process.”

⁵⁴ Smithies, Arthur, “Optimum Location in Spatial Competition,” Journal of Political Economy, 49 (06 1941), 423–439CrossRefGoogle Scholar, hypothesizes, on the basis of certain assumptions about turnout, that polarization of opinions could prevent convergence. Davis, Otto A. and Hinich, Melvin, “A Mathematical Model of Policy Formation in a Democratic Society,” in Mathematical Applications in Political Science, ed. Bernd, Joseph L. (Dallas: S.M.U. Press, 1966), pp. 175–208Google Scholar, suggests that the need to win nomination could lead to divergent policies on party-related issues.

We are presently completing a study of the positions which presidential candidates have taken on a number of policy issues. Our preliminary findings indicate that it is possible to specify conditions under which Republican and Democratic nominees do and do not tend to take similar positions.

⁵⁵ We are defining policy voting as a process in which an individual's policy preference, together with his perception of the candidates' stands on the issue, influences his vote. Under this definition it is possible for citizens with varying perceptions—even mispercep-tions—to engage in policy voting. (See footnote 13, above). The crucial question is whether opinions and perceptions are causally prior to the vote.

We present evidence, below, that although those who saw a big difference between Nixon and Humphrey tended to have congruent opinions, perceptions, and votes, much of the congruence resulted from projection rather than policy voting: vote intentions (as determined by party identification) were causally prior to perceptions.

⁵⁶ The greater strength of the relationship for Nixon may in part reflect his greater ambiguity on Vietnam, which made projection easier; but it also results from the fact that Republicans tended to like Nixon more than Democrats liked Humphrey. Republicans therefore felt a greater need to make their perceptions of Nixon's position congruent with their own opinions.

⁵⁷ Negative projection of the opposing candidate's position was more marked among Republicans than Democrats. Only 5 per cent of these Republicans saw Humphrey standing at exactly the same point as themselves, compared to the 14 per cent which would have been expected on the basis of the marginals. The overall negative relationship was significant at better than p = .001, by a chi-square test. The corresponding re-lationship for Democrats' perceptions of Nixon was not significant at the p = .10 level, however. The reason for this difference appears to be that Republicans felt negative toward Humphrey, but Democrats were merely neutral about Nixon and had less reason to project his position away.

All these tables are based only on voters who saw a large difference between the candidates. Projection was also widespread among other groups, such as non-voters, but it was much less common among voters who saw little difference between Nixon and Humphrey. We are grateful to Sue Bessmer for assistance in the analysis of perceptions.

⁵⁸ NYT, March 5, 1968, p. 1.

⁵⁹ “Vietnam,” August 1, 1968.

⁶⁰ NYT, March 11, 1968, p. 1.

⁶¹ NYT, March 31, 1968, p. 39.

⁶² NYT, April 2, 1968, p. 1.

⁶³ NYT, April 20, 1968, p. 17.

⁶⁴ NYT, May 4, 1968, p. 18.

⁶⁵ Most notably at the UPI Conference on October 7, where, in effect, he responded to Humphrey's Salt Lake City speech.

⁶⁶ Statement October 25, 1968, p. 2.

⁶⁷ “I See a Day …,” Nixon Speaks Out (New York: Nixon-Agnew Campaign Committee, 1968), pp. 277–289Google Scholar.

⁶⁸ This count is a generous one. It includes such allusions to Vietnam as, “We see Americans dying on distant battlefields abroad,” p. 278; and such Vietnam-related matters of “prevent[ing] more Vietnams,” p. 281, by asking “other nations in the Free World to bear their fair share of the burden of defending peace and freedom around this world,” p. 281, 282.

⁶⁹ Again, the count is generous; it includes sentences disavowing any promise to “eliminate all danger of war,” p. 280; promising “action—a new policy for peace abroad,” p. 280; declaring the policy of preventing “more Vietnams,” p. 281; and announcing that the time has come for other nations to bear more burdens of defending peace and freedom, p. 282.

⁷⁰ “I See a Day…,” p. 281.

⁷¹ “I See a Day…,” p. 278.

⁷² “I See a Day…,” p. 280.

⁷³ “I See a Day…,” p. 281.

⁷⁴ This was the Salt Lake City speech referred to above.

⁷⁵ NYT, April 23, 1968, p. 1; June 5, 1968, p. 34.

⁷⁶ NYT, May 27, 1968, p. 28; July 8, 1968, p. 31; October 17, 1968, p. 39; October 27, 1968.

⁷⁷ “A New Day for America,” Chicago, August 29, 1968.

⁷⁸ “A New Day for America,” p. 4.

⁷⁹ “A New Day for America,” p. 5.

⁸⁰ “A New Day for America,” p. 5.

⁸¹ See, for example, the interview in NYT, June 23, 1968, p. 58.

⁸² NYT, May 18, 1968, p. 1.

⁸³ NYT June 19, 1968.

⁸⁴ NYT, July 19, 1968, p. 18.

⁸⁵ NYT, July 20, 1968, p. 12; August 20, 1968; p. 1; August 27, 1968, p. 1; August 28, 1968, p. 30.

⁸⁶ NYT, September 9, 1968, p. 46.

⁸⁷ NYT, September 10, 1968, p. 32; September 11, 1968, p. 1, 28; September 12, 1968, p. 36.

⁸⁸ Similar reasoning is given in Downs, An Economic Theory of Democracy, pp. 135–137. Candidate ambiguity is predicted under special circumstances in Shepsle, “The Strategy of Ambiguity.”

⁸⁹ Kelley, Stanley, Political Campaigning, pp. 50–84Google Scholar, found that Eisenhower and Stevenson were vague concerning practically all issues in 1956. The televised debates of 1960 compelled Nixon and Kennedy to be somewhat more specific. Kelley, Stanley Jr., “The Presidential Campaign,” in David, Paul T., ed., The Presidential Election and Transition 1960–61 (Washington, D.C.: Brookings, 1961), p. 87Google Scholar.

We have found that Humphrey and Nixon took ambiguous positions on most policies in 1968. Nixon, especially, was quite ingenious at appearing to say much while actually saying little.

⁹⁰ A complete set of McCarthy's speeches was generously made available to us by Robert Axelrod and John D. Shilling. Sally Soth has helped greatly in analyzing these speeches.

⁹¹ Television interview on “Issues and Answers,” January 7, 1968, p. 14. This and subsequent page numbers refer—unless otherwise noted—to official press releases and transcripts.

⁹² Remarks, Stanford University, January 15, 1968, p. 4.

⁹³ Televised news confeernce, June 2, 1968, p. 6.

⁹⁴ Vietnam platform proposal, NYT, August 18, 1968, p. 66.

⁹⁵ Press conference, Indiana University, April 18, 1968, p. 4; televised interview on “Meet the Press,” July 7, 1968, p. 2.

⁹⁶ Press conference, Washington, D.C., June 12, 1968, p. 7.

⁹⁷ Remarks, San Francisco, May 29, 1968, p. 5.

⁹⁸ Response to question, Stanford University, January 15, 1968, pp. 3–4.

⁹⁹ Vietnam platform proposal, p. 66.

¹⁰⁰ Response to question, Wisconsin State University at Eau Claire, February 21, 1968, p. 12; press conference, Indiana University, April 18, 1968, p. 3.

¹⁰¹ Remarks, San Francisco, May 29, 1968, p. 5.

¹⁰² Response to question, University of Oregon, May 24, 1968.

¹⁰³ Vietnam platform proposal, p. 66.

¹⁰⁴ NYT, January 24, 1968, p. 55.

¹⁰⁵ NYT, March 4, 1968, p. 16.

¹⁰⁶ NYT, May 24, 1968, p. 20.

¹⁰⁷ NYT, May 27, 1968, p. 30.

¹⁰⁸ NYT, August 1, 1968, p. 20.

¹⁰⁹ In the seven months from January to the Republican convention, for example, the New York Times reported Reagan's comments on Vietnam only nine times. All of these reports were very brief and none were prominently featured.

¹¹⁰ Wallace, even more than other candidates, relied upon a single basic speech, which he modified only slightly as the campaign progressed. We have transcribed a tape recording distributed by the American Independent Party of a typical speech which Wallace delivered in Charlotte, North Carolina, in mid-June, 1968. We have relied upon the New York Times for reports of deviations from the basic speech.

¹¹¹ NYT, July 5, 1968, p. 14; July 27, 1968, p. 10; July 30, 1968, p. 27.

¹¹² Remarks, Charlotte, North Carolina, June 1968.

¹¹³ Remarks, Charlotte, N.C.

¹¹⁴ NYT, September 2, 1968, p. 17.

¹¹⁵ NYT, September 26, 1968, p. 53.

¹¹⁶ NYT, October 21, 1968, p. 38.

¹¹⁷ The choice of LeMay was very likely a mistake, which undercut Wallace's careful effort to be acceptable to those who opposed the war. LeMay disclaimed any intention of using nuclear weapons in Vietnam, NYT, October 4, 1968, pp. 50 ff.; but his hawkish image persisted.

¹¹⁸ The judges were 22 journalists, international relations specialists, and economists, from a variety of geographical areas. They gave McCarthy a median rating of approximately 2.0, and Wallace a rating of 6.6. They located Humphrey and Nixon much closer together and nearer the middle of the scale, at 4.0 and 5.0, respectively.

¹¹⁹ In August, 4 per cent rated McCarthy as an extreme hawk, and 6 per cent rated Wallace as an extreme dove. The percentages are based on those respondents, some 70 per cent of the sample, who made a rating.

¹²⁰ Slightly more than half the respondents placed him at the most extremely hawkish position on the scale.

¹²¹ In both the SRC and ORC/IEP November surveys, respondents were asked to rate the positions only of the active candidates and President Johnson.

¹²² The percentages, of those who rated both candidates, were 47 per cent and 11 per cent, respectively. Forty-two per cent saw no difference between the two, however, and in addition a large number of respondents failed to assign one or both candidates to any position on the scale.

As we have seen, public perceptions of Wallace's position were clearer in November. Indirect evidence suggests that perceptions of McCarthy improved as well, after the Democratic convention.

¹²³ These evaluations were measured on the 100-point SRC thermometer. See Weisberg and Rusk, “Dimensions of Candidate Evaluation.”

¹²⁴ Actual votes for Humphrey and Nixon correspond almost perfectly with comparative evaluations of the candidates. Brody, et al., “Vietnam, the Urban Crisis and the 1968 Presidential Election,” pp. 7–8Google Scholar.

¹²⁵ The slight nonlinearity in Table 5 suggests that a few respondents who placed themselves at the “immediate withdrawal” end of the scale may actually have favored a policy of “win or get out,” and supported Wallace.

¹²⁶ Vietnam preferences can account for approximately 4 per cent of the variance. The resemblance between the McCarthy-Reagan mock election and a real election is still not perfect, of course. If McCarthy and Reagan had been nominated, party identification might have affected evaluations of them more strongly, and might have blunted the impact of Vietnam opinions somewhat. On the other hand, nomination would have increased public exposure to the candidates' positions on Vietnam, and might on balance have increased the effect of Vietnam preferences.

¹²⁷ Key, V. O. Jr., The Responsible Electorate, pp. 2, 7Google Scholar.

¹²⁸ When we consider a single issue, upon which preferences are jointly single-peaked, a policy at the median of public opinion is the only policy which would not be defeated by any other in a series of two-way votes. The median may therefore be said to represent what “the public wants.”

As Figure 1 shows, the public generally perceived both Nixon and Humphrey as standing very close to the median of public opinion on the seven-point Vietnam scale. Commercial poll data and the candidates' speeches confirm that both candidates stood close to the median of public opinion on specific aspects of Vietnam policy.